Upload
michi
View
213
Download
0
Embed Size (px)
Citation preview
ARTICLE IN PRESSJID: EOR [m5G;October 21, 2014;15:23]
European Journal of Operational Research 000 (2014) 1–14
Contents lists available at ScienceDirect
European Journal of Operational Research
journal homepage: www.elsevier.com/locate/ejor
Innovative Applications of O.R.
Investment timing, debt structure, and financing constraints
Takashi Shibata a,∗, Michi Nishihara b,c
a Graduate School of Social Sciences, Tokyo Metropolitan University, 1-1 Minami-osawa, Hachioji, Tokyo 192-0397, Japanb Graduate School of Economics, Osaka University, 1-7 Machikaneyama, Toyonaka, Osaka 560-0043, Japanc Swiss Finance Institute, École Polytechnique Fédérale de Lausanne, 1015 Lausanne, Switzerland
a r t i c l e i n f o
Article history:
Received 27 June 2013
Accepted 1 September 2014
Available online xxx
Keywords:
Finance
Investment strategies
Real options
Debt structure
Debt issuance limit constraints
a b s t r a c t
We introduce debt issuance limit constraints along with market debt and bank debt to consider how finan-
cial frictions affect investment, financing, and debt structure strategies. Our model provides four important
results. First, a firm is more likely to issue market debt than bank debt when its debt issuance limit in-
creases. Second, investment strategies are nonmonotonic with respect to debt issuance limits. Third, debt
issuance limits distort the relationship between a firm’s equity value and investment strategy. Finally, debt
issuance limit constraints lead to debt holders experiencing low risk and low returns. That is, the more severe
the debt issuance limits are, the lower are the credit spreads and default probabilities. Our theoretical results
are consistent with stylized facts and empirical results.
© 2014 Elsevier B.V. All rights reserved.
1
(
b
r
l
g
h
fi
a
c
F
s
i
t
s
2
i
i
(
(
S
L
(
c
(
i
i
n
o
o
p
i
i
n
s
n
s
b
a
e
2 See, e.g., Stiglitz and Weiss (1981), Gale and Hellwig (1985), and Greenwald and
Stiglitz (1993) for static models of the investment decision under a financing constraint.
h
0
. Introduction
In the frictionless financial markets of Modigliani and Miller
1958), investment and financing decisions are completely separa-
le. However, when we introduce financial frictions such as corpo-
ate taxes, bankruptcy costs, and financing constraints, there is no
onger separability between a firm’s investment and financing strate-
ies. Since their seminal study, the corporate finance literature has
ighlighted the role of financial frictions between investment and
nancing decisions.1
Brennan and Schwartz (1984) and Mauer and Triantis (1994) ex-
mine the interaction between investment and financing decisions in
ontingent claim models. These models have two major limitations.
irst, there are no financial frictions. Second, the firm is financed by a
ingle type of debt, not various debt structures. The drawback of treat-
ng corporate debt as uniform is highlighted by the fact that different
ypes of debt instruments have quite different effects on investment
trategies (see Bolton & Freixas, 2000; Hackbarth, Hennessy, & Leland,
007; Rauh & Sufi, 2010).
Several recent studies have already begun the task of incorporat-
ng either financing frictions or various debt structures separately into
nvestment timing decision (real options) models. Boyle and Guthrie
2003), Hirth and Uhrig-Homburg (2010), and Nishihara and Shibata
∗ Corresponding author. Tel.: +81 42 677 2310; fax: +81 42 677 2298.
E-mail addresses: [email protected] (T. Shibata), [email protected]
M. Nishihara).1 A partial list includes Fazzari, Hubbard, and Petersen (1988), Hoshi, Kashyap, and
charfstein (1991), Kaplan and Zingales (1997), Hennessy and Whited (2007), and
ivdan, Sapriza, and Zhang (2009).
N
e
r
a
m
c
ttp://dx.doi.org/10.1016/j.ejor.2014.09.011
377-2217/© 2014 Elsevier B.V. All rights reserved.
Please cite this article as: T. Shibata, M. Nishihara, Investment timing
Operational Research (2014), http://dx.doi.org/10.1016/j.ejor.2014.09.011
2013) examine investment timing decisions under internal financing
onstraints. Nishihara and Shibata (2010) and Shibata and Nishihara
2012) investigate investment timing strategies under debt financ-
ng constraints.2 An interesting result among these earlier papers
s that investment strategies are nonmonotonic with respect to fi-
ancial frictions.3 Alternatively, most models consider a firm with
nly market debt (a single type of debt), assuming that dispersion
f creditors prevents debt reorganization during financial distress. In
ractice, a leveraged firm in financial distress can try to restructure
ts outstanding debt into a more affordable form. A model allow-
ng for debt restructuring approximates a firm with bank debt (i.e.,
onmarket debt).4 Sundaresan and Wang (2007) derive investment
trategies under various debt structures by considering debt reorga-
ization strategies for a firm under financial distress. However, these
trategies are derived independently of financing frictions.
In this paper, we assume that a firm can issue two classes of debt:
ank and market debt. Following Bulow and Shoven (1978), Gertner
nd Scharfstein (1991), and Hackbarth et al. (2007), the only differ-
nce between bank and market debt is the bankruptcy procedure.5
3 See Boyle and Guthrie (2003), Hirth and Uhrig-Homburg (2010), and Shibata and
ishihara (2012) for theoretical analyses. See Cleary, Povel, and Raith (2007) for an
mpirical analysis.4 A partial list of structural models with bank debt includes Mella-Barral and Per-
audin (1997), Fan and Sundaresan (2000), Broadie, Chernov, and Sundaresan (2007),
nd Hackbarth et al. (2007). None of these papers considers investment strategies.5 They assume that payments to market lenders cannot be changed outside the for-
al bankruptcy process and that the new owners can recapitalize optimally, although
osts are incurred.
, debt structure, and financing constraints, European Journal of
2 T. Shibata, M. Nishihara / European Journal of Operational Research 000 (2014) 1–14
ARTICLE IN PRESSJID: EOR [m5G;October 21, 2014;15:23]
a
f
&
G
c
i
e
a
s
e
v
t
m
t
l
i
s
v
i
d
i
a
t
(
v
d
r
t
u
T
d
t
a
i
a
r
p
s
a
l
a
k
c
e
s
t
c
2
t
a
o
m
d
t
Bank loan renegotiations are generally easier than bond restructur-
ing, as suggested in Lummer and McConnell (1989) and Gilson, John,
and Lang (1990). In addition, we assume that firms face issuance lim-
its for debt. It is widely observed that firms face debt issuance limits
in financing investment. Debt issuance limit constraints are typically
needed to rule out corporate default by mitigating risk shifting from
equity holders to debt holders (see Jensen & Meckling, 1976). Thus,
imposing debt issuance limit constraints in economic models can thus
capture an important feature of reality.
We introduce debt issuance limit constraints along with market
debt and bank debt to consider how financial frictions affect invest-
ment, financing, and debt structure strategies. Our model endoge-
nously determines investment timing, coupon payment level, and
debt structure (either bank or market debt issuance) in the presence
of financial frictions. To be more precise, our model is solved as fol-
lows. Given a debt structure, we derive the optimal investment and
coupon payment strategies under debt issuance limit constraints. We
then choose the optimal debt structure by comparing equity value
under bank debt with that under market debt.
Our model builds primarily on three papers: McDonald and Siegel
(1986), Sundaresan and Wang (2007), and Shibata and Nishihara
(2012).6 Our model becomes an all-equity financing model when the
firm cannot issue any type of debt (i.e., McDonald & Siegel, 1986).
Our model is a nonconstrained model under various debt structures
when the firm can issue bank and market debt without issuance con-
straints (i.e., Sundaresan & Wang, 2007). Our model becomes a con-
strained model under a market debt structure when a firm can issue
only market debt with an issuance bound constraint (i.e., Shibata
& Nishihara, 2012). Our model can be regarded as a natural exten-
sion of these three papers, and yields several additional important
implications.
Our model provides four important results. The first result is
that the type of debt that is issued at the time of investment de-
pends largely on three parameters (debt issuance limit, volatility, and
firm’s bargaining power during financial distress). The firm is more
likely to issue market debt than bank debt when the debt issuance
limit is increased.7 In practice, the firms with higher (lower) debt
issuance bounds are regarded as large/mature (small/young) corpo-
rations. Based on this definition, our results show that large/mature
(small/young) corporations are more likely to choose market (bank)
debt. Thus, this segmentation is broadly consistent with the stylized
facts. In addition, we consider the effects of volatility and bargaining
power. When volatility is sufficiently high, the firm prefers bank debt
financing irrespective of bargaining power. When volatility is high
and the firm’s bargaining power is low, the firm prefers bank debt
financing. The reason is that higher volatility is more likely to lead to
default, and a lower firm’s bargaining power is more likely to lead to
bank debt issuance. When volatility is low and the firm’s bargaining
power is high, the firm prefers market debt financing. This is because
lower volatility is less likely to result in default, and a higher firm’s
bargaining power causes less bank debt issuance. These findings fit
well with the empirical findings of Blackwell and Kidwell (1988) and
Denis and Mihov (2003).
The second result is that the investment strategies (one of two
solutions) are nonmonotonic with respect to the debt issuance con-
straints, while the coupon payments (the other solution) are mono-
tonic with respect to the debt issuance constraints. Given a debt struc-
ture, the investment thresholds have a U-shaped relationship with
the debt issuance constraints. The U-shaped relationship under bank
debt financing is a new result, although the U-shaped relationship
under market debt financing has already been established in Shibata
6 We incorporate the investment decision into a financing decision model having
the choice of debt structure developed by Hackbarth et al. (2007).7 This result is similar to that of Hackbarth et al. (2007) who do not consider invest-
ment strategies.
d
b
fi
c
&
Please cite this article as: T. Shibata, M. Nishihara, Investment timin
Operational Research (2014), http://dx.doi.org/10.1016/j.ejor.2014.09.011
nd Nishihara (2012). The nonmonotonicity between investment and
rictions is similar to that in previous related papers (see, e.g., Boyle
Guthrie, 2003; Cleary et al., 2007; Hirth & Uhrig-Homburg, 2010).
iven a debt structure, the coupon payments are monotonically in-
reasing with respect to the debt issuance constraints. These results
mply that the optimal investment thresholds of the constrained lev-
red firm are not in between those of the unconstrained levered firm
nd unlevered firm, while the optimal coupon payments of the con-
trained levered firm are in between those of the unconstrained lev-
red firm and unlevered firm. Thus, it is less costly to distort in-
estment thresholds than to distort coupon payments. Moreover, if
he optimal debt structure strategies are changed from bank debt to
arket debt by increasing the debt issuance bound, the investment
hresholds and coupon payments have a discontinuous curved re-
ationship with respect to the debt issuance limit. In particular, the
nvestment thresholds may have a discontinuous W-shaped relation-
hip with the debt issuance limit.
The third result is about the relationship between equity option
alues and its investment thresholds. Suppose, as a benchmark, there
s no debt issuance limit constraint. Then, if the firm prefers bank
ebt financing, the investment thresholds under bank debt financ-
ng are lower than those under market debt financing. We call this
“symmetric relationship.” This implies that having the opportunity
o choose the debt structure always hastens corporate investment
decreases the investment threshold) and increases its equity options
alue. Suppose, by contrast, the firm is financially constrained by its
ebt issuance limit. Then we show that there is not always a symmet-
ic relationship between them. To be more precise, the investment
hresholds under bank debt financing are not always lower than those
nder market debt financing, even when the firm prefers bank debt.
his implies that having the opportunity to choose a debt structure
oes not always hasten corporate investment although its equity op-
ion value is increased. Thus, we show that financing constraints cause
distortion to the symmetric relationship between investment and
ts equity value.
The final result is that debt issuance constraints create a low-risk
nd low-return outcome for debt holders. In our model, return and
isk for debt holders can be measured by the credit spreads and default
robabilities, respectively. We show that, the more severe the debt is-
uance bounds are, the lower are the credit spreads and default prob-
bilities. Thus, we shed light on the determinants of the debt issuance
imits for debt holders. Moreover, we show that the credit spreads
nd default probabilities for bank debt are higher than those for mar-
et debt. These imply that bank debt is high risk and high return
ompared with market debt. These results are consistent with the
mpirical results of Davydenko and Strebulaev (2007).
The remainder of the paper is organized as follows. Section 2 de-
cribes the model and derives the value functions. Section 3 provides
he solution of our model and considers its properties. Section 4 dis-
usses the model’s implications. Section 5 concludes.
. Model
In this section, we begin with a description of the model. We
hen provide the value functions for the firms financed by bank debt
nd by market debt. Then, we formulate our model as an investment
ptimization problem for the firms financed by bank debt and by
arket debt with issuance limit constraints. As a benchmark, we
erive the solutions to the two extreme cases in our model. One is
he solution for an unlevered (all-equity financed) firm when the
ebt issuance limit constraints are strict (i.e., the model developed
y McDonald & Siegel, 1986). The other is the solution for a firm
nanced by bank debt and market debt when the debt issuance limit
onstraints are immaterial (i.e., a model similar to that of Sundaresan
Wang, 2007).
g, debt structure, and financing constraints, European Journal of
T. Shibata, M. Nishihara / European Journal of Operational Research 000 (2014) 1–14 3
ARTICLE IN PRESSJID: EOR [m5G;October 21, 2014;15:23]
0time
cash
inflo
w
x1i
x1d
T1i T
1d
region a
0time
cash
inflo
w
T2d
x2d
x2i
region a
T2d
region b
T2i
Fig. 1. Difference between market debt and bank debt.
2
t
“
a
w
d
w
μw
r
l
p
(
c
i
H
b
p
U
l
U
a
t
b
m
T
v
r
n
N
a
i
t
b
d
c
t
fi
t
w
(
f
T
b
m
b
x
t
m
c
i
f
r
m
c
i
l
2
d
t
.1. Setup
A firm possesses an option to invest in a single project at any
ime. If the investment option is exercised at time T i (superscript
i” indicates investment strategies), the firm pays a fixed cost I > 0
t time T i and receives an instantaneous cash flow Xt after time T i,
here Xt is given by the following geometric Brownian motion:
Xt = μXtdt + σXtdzt, X0 = x, (1)
here zt denotes a standard Brownian motion, and the growth rate
> 0 and volatility σ > 0 are constant parameters. For convergence,
e assume that r > μ > 0,8 where r is a constant risk-free interest
ate. It is assumed that the current state variable X0 = x is sufficiently
ow so that the investment is not undertaken immediately.
In this paper, we assume that the firm issues two classes of per-
etual debt: market debt with a promised coupon payment flow c1
subscript “1” indicates market debt) and bank debt with a promised
oupon payment flow c2 (subscript “2” indicates bank debt). Follow-
ng Bulow and Shoven (1978), Gertner and Scharfstein (1991), and
ackbarth et al. (2007), we assume that the only difference between
ank debt and market debt is the bankruptcy procedure. Now sup-
ose that the firm in financial distress tries to restructure the debt.
nder market debt financing, the coupon payments to the market
ender cannot be changed outside the formal bankruptcy process.
nder bank debt financing, the coupon payments to the bank lender
re reduced in the course of a costless private workout.
Given a debt structure j ( j ∈ {1, 2}), let us denote by T ij
and Tdj
he investment (indicated by superscript “i” ) and default (indicated
y superscript “d”) timings, respectively. Mathematically, the invest-
ent and default timings are defined as T ij= inf{s ≥ 0 | Xs = xi
j} and
dj
= inf{s ≥ T ij| Xs = xd
j}, where xi
jand xd
jdenote the associated in-
estment and default thresholds, respectively. In particular, xd1 and xd
2
epresent the formal bankruptcy threshold for market debt and the
egotiation (coupon reduction) threshold for bank debt, respectively.
ote that these defaults are defined only after issuing the debt at the
8 The assumption r > μ ensures that the value of the firm is finite. See Sundaresan
nd Wang (2007), Kort, Murto, and Pawlina (2010), and Shibata and Nishihara (2011)
n detail.
“
Please cite this article as: T. Shibata, M. Nishihara, Investment timing
Operational Research (2014), http://dx.doi.org/10.1016/j.ejor.2014.09.011
ime of investment in this model (i.e., the firm is an unlevered firm
efore investment).
Fig. 1 illustrates the difference between market debt and bank
ebt. On the one hand, the left-hand side panel demonstrates the
ase of market debt financing. Once Xt , starting at X0 = x, arrives at
he threshold xi1, where superscript “i” indicates the investment, the
rm exercises the investment by issuing the market debt. If Xt hits
he formal bankruptcy threshold xd1 after adopting the investment,
here superscript “d” indicates the default, the firm is liquidated
the firm stops the operation). Mathematically, the investment and
ormal bankruptcy timings are defined as T i1 = inf{t ≥ 0 | Xt ≥ xi
1} andd1 = inf{t ≥ T i
1 | Xt ≤ xd1}.
On the other hand, the right-hand side panel depicts the case of
ank debt financing. Once Xt , starting at X0 = x, arrives at the invest-
ent threshold xi2, the firm exercises the investment by issuing the
ank debt. The investment timing is defined as T i2 = inf{t ≥ 0 | Xt ≥
i2}. If Xt hits the coupon switching threshold xd
2 from the above point,
he normal coupon payment is changed to the reduced coupon pay-
ent. In addition, if Xt hits the same threshold xd2 from below, the
oupon payment is changed to the normal coupon payment. The tim-
ngs for coupon switching are defined by Td2 := inf{t ≥ T i
2 | Xt = xd2},
rom the normal coupon to the reduced coupons as well as from the
educed coupon to the normal coupon.9 Thus, the firm pays the nor-
al coupon in the region Xt ≥ xd2, while the firm pays the reduced
oupon in the region Xt < xd2.10 Because the reduced coupon payment
n the region Xt < xd2 depends on Xt at the equilibrium, we can avoid
iquidation under bank debt financing.
.2. Value functions of the firm financed by market debt
In this subsection, we derive the value functions after the market
ebt is issued at the time of investment.
Let us denote by Ea1(Xt, c1) the equity value at time t after issuing
he market debt at the time of investment, where the superscript
a” represents firm value after the market debt issuance. The value,
9 For notational simplicity, we do not distinguish Td2 from above and below.
10 See Dixit (1989) for the switching model.
, debt structure, and financing constraints, European Journal of
4 T. Shibata, M. Nishihara / European Journal of Operational Research 000 (2014) 1–14
ARTICLE IN PRESSJID: EOR [m5G;October 21, 2014;15:23]
E
D
H
x
a
D
w
T
(
p
r
s
V
T
V
w
t
V
r
d
s
T
l
m
b
fi
o
θ
2
Ea1(Xt, c1), is defined as
Ea1(Xt, c1) = sup
Td1
Et
[ ∫ Td1
t
e−r(u−t)(1 − τ)(Xu − c1)du
], (2)
where Et denotes the expectation operator at time t, and τ denotes
the tax rate. Note that Td1 is the formal bankruptcy time that the equity
holders optimize. As in standard arguments, Ea1(Xt, c1) is given by
a1(Xt, c1)
= maxxd
1
{�Xt − (1 − τ)
c1
r−
(�xd
1 − (1 − τ)c1
r
)(Xt
xd1
)γ }, (3)
where � := (1 − τ)/(r − μ) > 0 and γ := 1/2 − μ/σ 2 − ((μ/σ 2 −1/2)2 + 2r/σ 2)1/2 < 0. Then, the optimal threshold for formal
bankruptcy is obtained by
xd1(c1) = argmax
xd1
Ea1(Xt, c1) = κ−1
1 c1, (4)
where κ1 := (γ − 1)�r/(γ (1 − τ)) > 0. Note that xd1(c1) is given as a
linear function of c1 with limc1↓0 xd1(c1) = 0. The result is obtained by
Black and Cox (1976).
The debt value, Da1(Xt, c1), is given as
Da1(Xt, c1) = Et
[ ∫ Td1
t
e−r(u−t)c1du + e−r(Td1 −t)(1 − α)�xd
1(c1)
]
= c1
r+
((1 − α)�xd
1(c1)− c1
r
)(Xt
xd1(c1)
)γ
, (5)
where α ∈ (0, 1) denotes the proportional cost of formal bankruptcy.
Because limc1↓0 xd1(c1) = 0 in (4), we have limc1↓0 Da
1(Xt, c1) = 0. The
total firm value is defined by the sum of the equity and debt values,
Va1(Xt, c1) := Ea
1(Xt, c1)+ Da1(Xt, c1), i.e.,
Va1(Xt, c1) = �Xt + τ c1
r
(1 −
(Xt
xd1(c1)
)γ )− α�xd
1(c1)
(Xt
xd1(c1)
)γ
.
(6)
There are three components of the right-hand side in (6). The first,
second, and third terms represent the values of the unlevered firm, the
tax shield, and the formal bankruptcy cost, respectively. Obviously,
we have limc1↓0 Va1(Xt, c1) = �Xt .
2.3. Value functions for the firm financed by bank debt
This subsection provides the value functions after the bank debt
is issued at the time of investment.
When the bank debt is issued at the time of investment, there are
two types of regions: normal and renegotiation regions. On the one
hand, let us denote by Ea2(Xt, c2), Da
2(Xt, c2), and Va2(Xt, c2) the equity,
debt, and total firm values, respectively, in the normal region, where
the superscript “a” stands for the normal region after issuing the
bank debt. On the other hand, let us denote by Eb2(Xt, c2), Db
2(Xt, c2),
and Vb2 (Xt, c2) the equity, debt, and total firm values, respectively,
in the negotiation region during the period of financial distress. The
superscript “b” indicates the negotiation (coupon reduction). The nor-
mal and negotiation regions are divided by the negotiation (coupon
reduction) threshold xd2. That is, the region {Xt ≥ xd
2} is the normal
region, while the region {Xt < xd2} is the negotiation region. The firm
and bank negotiate and divide the surplus Vb2 (Xt, c2)− (1 − α)�Xt to
avoid formal bankruptcy. The division of the surplus between the firm
and the bank depends on their relative bargaining powers. Let us de-
note by η and 1 − η the bargaining powers of the firm and the bank,
respectively (η ∈ [0, 1]).
The equity and debt values in the normal region, Ea2(Xt, c2) and
Da2(Xt, c2), respectively, are given by
Ea2(Xt, c2) = sup
Td2 ≥t
Et
[ ∫ Td2
t
e−r(u−t)(1 − τ)(Xu − c2)du
Please cite this article as: T. Shibata, M. Nishihara, Investment timin
Operational Research (2014), http://dx.doi.org/10.1016/j.ejor.2014.09.011
+ e−r(Td2 −t)Eb
2(XTd2, c2)
], (7)
a2(Xt, c2) = Et
[ ∫ Td2
t
e−r(u−t)c2du + e−r(Td2 −t)Db
2(XTd2, c2)
]. (8)
ere, the coupon switching times are defined by Td2 = inf{t ≥ 0, Xt =
d2}. The equity and debt values in the negotiation region, Eb
2(Xt, c2)
nd Db2(Xt, c2), respectively, are given by
Eb2(Xt, c2) = Et
[ ∫ Td2
t
e−r(u−t)((1 − τ)Xu − s(Xu))du
+ e−r(Td2 −t)Ea
2(XTd2, c2)
], (9)
b2(Xt, c2) = Et
[ ∫ Td2
t
e−r(u−t)s(Xu)du + e−r(Td2 −t)Da
2(XTd2, c2)
],
(10)
here s(Xt) is the reduced coupon payment in the negotiation region.
he total firm value is defined by Vl2(Xt, c2) = El
2(Xt, c2)+ Dl2(Xt, c2)
l ∈ {a, b}).
We begin by solving the total firm values. By the standard variation
rinciple, we have the following differential equations:
rVa2(x, c2) = (1 − τ)x + τ c2 + μx
∂Va2(x, c2)
∂x
+ 1
2σ 2x2 ∂2Va
2(x, c2)
∂x2, Xt ≥ xd
2, (11)
Vb2 (x, c2) = (1 − τ)x + μx
∂Vb2 (x, c2)
∂x
+ 1
2σ 2x2 ∂2Vb
2 (x, c2)
∂x2, Xt < xd
2, (12)
ubject to the boundary conditions
a2
(xd
2, c2
) = Vb2
(xd
2, c2
),
∂Va2(x, c2)
∂x
∣∣∣∣x=xd
2
= ∂Vb2 (x, c2)
∂x
∣∣∣∣x=xd
2
.
hus, Va2(x, c2) and Vb
2 (x, c2) are obtained by
Va2(Xt, c2) = �Xt + τ c2
r
(1 − β
β − γ
(Xt
xd2
)γ ), Xt ≥ xd
2, (13)
b2 (Xt, c2) = �Xt − τ c2
r
γ
β − γ
(Xt
xd2
)β
, Xt < xd2, (14)
here β := 1/2 − μ/σ 2 + ((μ/σ 2 − 1/2)2 + 2r/σ 2)1/2 > 1. Note that
here is no bankruptcy cost term of Va2(Xt, c2) in (13), in contrast with
a1(Xt, c1) in (6).
We then derive the reduced coupon payment in the negotiation
egion. As in the derivation in Sundaresan and Wang (2007), the re-
uced coupon payment in the renegotiation region, s(Xt), is given by
(Xt) = (1 − αη)(1 − τ)Xt, Xt < xd2. (15)
he reduced coupon payment s(x) is a linear function of x with
imx↓0 s(x) = 0. The lower the EBIT x is, the lower is the coupon pay-
ent s(x). This means that the firm does not need to consider formal
ankruptcy once it is in the negotiation region.
Let θe(Xt, c2) denote the fraction of firm value Vb2 (Xt, c2) that the
rm receives from renegotiation. Using Nash bargaining, θe(Xt, c2) is
btained by solving
e(Xt, c2) = argmaxθ
(θVb
2 (Xt, c2))η
× ((1 − θ)Vb
2 (Xt, c2)− (1 − α)�Xt
)1−η
= η − η(1 − α)�Xt
Vb(Xt, c2).
g, debt structure, and financing constraints, European Journal of
T. Shibata, M. Nishihara / European Journal of Operational Research 000 (2014) 1–14 5
ARTICLE IN PRESSJID: EOR [m5G;October 21, 2014;15:23]
T
D
w
E
E
T
∂
x
w
x
i
D
I
2
f
c
W
s
f
t
d
i
m
e
a
1
r
D
c
t
s
r
t
E
s
P
n
u
m
v
E
w
E
f
m
r
t
d
r
t
w
2
c
fi
l
(
E
f
v
x
w
2
T
n
s
T
E
w
E
f 1N 2N
11 Hackbarth et al. (2007) consider the mixed debt policy because they incorporate
the priority structures of bank and market debt. Note that we do not introduce them.12 This problem is equivalent to the investment problem for an all-equity financed
firm, which is the simple version of the seminal model by McDonald and Siegel (1986).13 This problem is similar to that in Sundaresan and Wang (2007).
hus, the sharing rules are
Eb2(Xt, c2) = ηVb
2 (Xt, c2)− η(1 − α)�Xt
= η
(α�Xt − τ c2
r
γ
β − γ
(Xt
xd2
)β), (16)
b2(Xt, c2) = (1 − η)Vb
2 (Xt, c2)+ η(1 − α)�Xt
= (1 − αη)�Xt − (1 − η)τ c2
r
γ
β − γ
(Xt
xd2
)β
, (17)
here Xt < xd2. Moreover, the equity value in the normal region,
a2(Xt, c2) in (7), is rewritten as
a2(Xt, c2) = max
xd2
{�Xt − (1 − τ)
c2
r
−((1 − αη)�xd
2 − c2
r
(1 − τ − τ
ηγ
β − γ
))(Xt
xd2
)γ }.
(18)
he optimal coupon switching threshold is chosen to satisfy
Ea2/∂xd
2 = ∂Eb2/∂xd
2, i.e.,
d2(c2) = κ−1
2 c2, (19)
here κ2 := (γ − 1)(1 − αη)�r/(γ (1 − τ(1 − η)) > 0. Note thatd2(c2) is a linear function of c2 with limc2↓0 xd
2(c2) = 0. The debt value
n the normal region, Da2(Xt, c2) in (8), is given by
a2(Xt, c2) = c2
r+
((1 − αη)�xd
2(c2)
− c2
r
(1 − τ + τ
β
β − γ− τ
ηγ
β − γ
))(Xt
xd2(c2)
)γ
. (20)
t is straightforward to obtain limc2↓0 Da2(Xt, c2) = 0.
.4. Investment problem for the constrained levered firm
In this subsection, we formulate the investment decision problem
or the firm financed by bank debt and by market debt with issuance
onstraints.
Suppose that the debt structure j is given ( j ∈ {1, 2}). Following
ong (2010) and Shibata and Nishihara (2012), the firm’s debt is-
uance limit constraint is assumed to be
Daj
(xi
j, cj
)I
≤ q, (21)
or some constant q ≥ 0. The inequality (21) means that the ratio of
he debt issuance amount to the investment cost is restricted by the
ebt issuance ratio q. In other words, the firm has access to debt
ssuance Daj(xi
j, cj) up to the amount qI ≥ 0. For example, the friction
ight arise because of the risk-shifting problem. Because issuing debt
ncourages risk shifting from equity holders to debt holders, investors
re reluctant to lend beyond a certain amount (see Jensen & Meckling,
976). Note that we have assumed q ∈ [0,+∞), not q ∈ [0, 1]. The
eason is that the levered firm has the possibility of issuing the amountaj(xi
j, cj), which is more than the amount I, to maximize its value. The
ase of Daj(xi
j, cj) > I may lead to the result q > 1, which implies that
he excess is distributed to equity holders as a dividend.
Let us denote by Eoj(x) the equity options value of the firm con-
trained by debt issuance limit ( j ∈ {1, 2}), where the superscript “o”
epresents the options value before investment. Given a debt struc-
ure j, the equity options value of the constrained firm is defined as
oj (x) := sup
T ij≥0,cj≥0
E0[e−rT i
j {Eaj (XT i
j, cj)− (
I − Daj (XT i
j, cj)
)}],
ubject to Daj(X
Tij, cj) ≤ qI. Using standard arguments in Dixit and
indyck (1994), we have E0[e−rTi
j ] = (x/xij)β where X0 = x < xi
j.
Please cite this article as: T. Shibata, M. Nishihara, Investment timing
Operational Research (2014), http://dx.doi.org/10.1016/j.ejor.2014.09.011
We formulate the investment decision problem of the firm fi-
anced by bank debt and market debt with issuance constraints. Let
s denote by Eo(x) the equity options value of the firm with opti-
al debt structure subject to issuance constraints. The equity options
alue Eo(x) is obtained by
o(x) = max {Eo1(x), Eo
2(x)}, (22)
here Eoj(x) is given by
oj (x) = max
xij≥0,cj≥0
(x
xij
)β
{Vaj
(xi
j, cj
) − I}, (23)
subject toDa
j
(xi
j, cj
)I
≤ q, j ∈ {1, 2}, (24)
or any x < min{xi1, xi
2}.
This problem is related to the issuance of either bank debt or
arket debt at the time of investment. The reason is that we can
emove the mixed issuance of bank debt and market debt (we show
hat this assumption is correct below).11
Before analyzing the optimal investment strategies for a firm with
ifferent debt structures subject to a debt issuance constraint, we first
eview two extreme cases briefly: the optimal investment problems of
he unlevered firm and the levered firm with different debt structures
ithout issuance constraints.
.5. Investment problem for the unlevered firm
In this subsection, we assume that q = 0, i.e., Daj(xi
j, 0) = 0 and
j = 0 for all j ( j ∈ {1, 2}).12
Let us denote by EoU(·) the equity options value of the unlevered
rm before investment, where the subscript “U” stands for the un-
evered firm. When q = 0, we have Eaj(xi
j, 0) = Va
j(xi
j, 0) = �xi
jfor all j
j ∈ {1, 2}). Thus, the value, EoU(x), turns out to be
oU(x) = max
xiU≥0
(x
xiU
)β
{�xiU − I}, (25)
or any x < xiU. The optimal investment strategy and the equity options
alue are
i∗U = β
β − 1
1
�I, Eo
U(x) =(
x
xi∗U
)β
(β − 1)−1I, (26)
here the superscript “∗” represents the optimum.
.6. Investment problem for the unconstrained levered firm
In this subsection, we assume that q is sufficiently large (q ↑ +∞).
hen, the debt issuance constraint becomes immaterial.13
Let us denote by EoN(x) the equity options value for the firm fi-
anced by bank debt and market debt without an issuance con-
traint, where the subscript “N” represents the nonconstrained firm.
he value, EoN(x), is formulated as
oN(x) = max {Eo
1N(x), Eo2N(x)}, (27)
here EojN
(x) is given by
ojN(x) = max
xijN
≥0,cjN≥0
(x
xijN
)β
{Vaj
(xi
jN, cjN
) − I}, (28)
or any x < min{xi , xi } ( j ∈ {1, 2}).
, debt structure, and financing constraints, European Journal of
6 T. Shibata, M. Nishihara / European Journal of Operational Research 000 (2014) 1–14
ARTICLE IN PRESSJID: EOR [m5G;October 21, 2014;15:23]
00
x
Eo (
equi
ty o
ptio
n va
lue)
EUo (x)
EUa(x)
EjNo (x)
Eja(x,c
jN)
xU∗
(β−1)−1I
xjN∗
Fig. 2. Equity option values and investment thresholds without financing constraints.
3
i
p
t
i
d
p
3
v
D
w
g
e
(
t
0
v
t
l
d
h
c
x
t
t
t
L
x
O
f
c
r
P
j
s
w
f
f
f
The solutions and equity option value for a given debt structure j
are14
xi∗jN = ψjx
i∗U , c∗
jN = κjψj
hj
xi∗U , Eo
jN(x) = ψ−βj
EoU(x), (29)
where
h1 :=(
1 − γ
(1 + α
1 − τ
τ
))−1/γ
≥ 1,
ψ1 :=(
1 + τ
1 − τ
1
h1
)−1
≤ 1,
h2 :=(
β
β − γ(1 − γ )
)−1/γ
≥ 1,
ψ2 :=(
1 + τ(1 − αη)
1 − τ(1 − η)
1
h2
)−1
≤ 1. (30)
We summarize these results obtained in (29) as follows.
Lemma 1. Suppose that the firm does not have a debt issuance constraint
(i.e., q is sufficiently large). The first property is that we have xi∗jN
≤ xi∗U and
EojN
(x) ≥ EoU(x) for any j ( j ∈ {1, 2}). The second property is that ψ1 < ψ2
implies that xi∗1N < xi∗
2N and Eo1N(x) > Eo
2N(x). The third property is that
EoU(xi∗
U ) = Eo1(x
i∗1N) = Eo
2(xi∗2N).
Lemma 1 is easily demonstrated by Fig. 2 which depicts the equity
option value with the state variable. The first property is the same
as that in the static model of Myers (1977). The second property
corresponds to the third property. From these two properties, we
have xi∗1N < xi∗
2N if and only if Eo1N(x) > Eo
2N(x). We call this property
symmetric relationship.
Before solving our problem, we consider the properties of the so-
lution and its value. We have the following two intuitive conjectures:
xi∗j ∈ [xi∗
jN, xi∗U ] for any q, xi∗
1 ≤ xi∗2 if and only if Eo
1(x) ≥ Eo2(x).
The first conjecture is implied by the fact that xi∗U = limq↓0 xi∗
j>
limq↑+∞ xi∗j
= xi∗jN
for any q ( j ∈ {1, 2}). The second one is implied by
the fact that xi∗1N ≤ xi∗
2N if and only if Eo1N(x) ≥ Eo
2N(x). However, inter-
estingly, these intuitive conjectures are not necessarily correct.
14 See Sundaresan and Wang (2007) and Shibata and Nishihara (2010) for a detailed
derivation.
f
Please cite this article as: T. Shibata, M. Nishihara, Investment timin
Operational Research (2014), http://dx.doi.org/10.1016/j.ejor.2014.09.011
. Model solution
In this section, we derive the solution to the problem formulated
n (22). Section 3.1 provides the solution to the investment decision
roblem for the constrained levered firm with a given debt struc-
ure. Section 3.2 examines the optimal debt structure by compar-
ng the equity options values for a firm with bank debt and market
ebt with issuance constraints. Then, we obtain the solution to the
roblem (22).
.1. Optimal investment strategies for a given debt structure
Given a debt structure j ( j ∈ {1, 2}), we define the constant critical
alue xpj
satisfying
aj
(x
pj, cj
(x
pj
)) = qI, (31)
here the superscript “p” represents the critical point and cj(x) is
iven as the optimal coupon payment of the nonconstrained lev-
red firm given a debt structure j,15 i.e., cj(x) := argmaxcjVa
j(x, cj) =
κj/hj)x as shown in Leland (1994). Then, Daj(x, cj(x))is a strictly mono-
onically increasing continuous function of x with limx↓0 Daj(x, cj(x)) =
and limx↑+∞ Daj(x, cj(x)) = +∞. Thus, there exists a unique critical
alue xpj
.
Suppose, for example, xi∗jN
> xpj
for any j ( j ∈ {1, 2}), implying that
he investment threshold of the nonconstrained levered firm is strictly
arger than the critical value. Then, the firm would prefer to issue more
ebt than the amount qI to maximize the equity value, while the debt
olders restrict the amount to less than qI. Thus, the firm is financially
onstrained by its debt issuance bound. Suppose that, instead, xi∗jN
≤pj
. Then, the firm can choose the investment threshold to maximize
he total firm value on the condition that the debt issuance is less than
he bound qI. Therefore, the firm is not constrained. We summarize
hese results as follows.
emma 2. Given a debt structure j ( j ∈ {1, 2}), there exists a uniquepj
satisfying (31). Then, if xi∗jN
> xpj
, the firm is financially constrained.
therwise, it is not.
Lemma 2 implies that, by comparing the magnitudes of xi∗jN
and xpj
or a given debt structure j, we recognize whether the debt issuance
onstraint is binding. Using these properties, we have the following
esults (the proof is given in Appendix A).
roposition 1. Suppose that xi∗jN
> xpj
≥ 0 for a given debt structure
(j ∈ {1, 2}). For xpj
= 0, we have xi∗j
= xi∗U and c∗
j= 0. For x
pj
> 0, the
olutions, xi∗j
and c∗j, are decided uniquely by two simultaneous equations:
fj1
(xi∗
j, c∗
j
)fj2
(xi∗
j, c∗
j
) −fj3
(xi∗
j, c∗
j
)fj4
(xi∗
j, c∗
j
) = 0, Daj (x
i∗j , c∗
j )− qI = 0, (32)
here fj1, fj2, fj3, and fj4 are given as
11 := (β − 1)�xi∗1
+(β
τ
r− (β − γ )
(κ1xi∗
1
c∗1
)γ (τ
r+ α�
κ1
))c∗
1 − βI, (33)
12 :=(
κ1xi∗1
c∗1
)γ
γ
(1
r− (1 − α)
�
κ1
)c∗
1, (34)
13 := τ
r−
(κ1xi∗
1
c∗1
)γ
(1 − γ )
(τ
r+ α�
κ1
), (35)
14 := 1
r−
(κ1xi∗
1
c∗
)γ
(1 − γ )
(1
r− (1 − α)�
κ1
), (36)
1
15 Note that we have cj(xi∗jN) = c∗
jN, where c∗
jNis defined by the second term of (29).
g, debt structure, and financing constraints, European Journal of
T. Shibata, M. Nishihara / European Journal of Operational Research 000 (2014) 1–14 7
ARTICLE IN PRESSJID: EOR [m5G;October 21, 2014;15:23]
a
f
f
f
f
S
o
v
E
3
v
e
o
o
p
τi
(
E
E
t
o
a
n
E
a
v
O
l
i
w
F
w
c
p
p
w
fi
c
v
{
l
o
p
l
a
E
r
O
l
d
F
b
o
a
s
l
d
w
I
w
t
a
w
4
o
c
s
t
w
d
c
d
h
4
o
w
h
C
l
i
t
(
f
t
h
nd
21 := (β − 1)�xi∗2 + β
τ
r
(1 −
(κ2xi∗
2
c∗2
)γ )c∗
2 − βI, (37)
22 :=(
κ2xi∗2
c∗2
)γ
×γ
(1
r
(1 − τ + τ
β
β − γ− τ
ηγ
β − γ
)− (1 − αη)
�
κ2
)c∗
2,
(38)
23 := τ
r
(1 −
(κ2xi∗
2
c∗2
)γ
(1 − γ )β
β − γ
), (39)
24 := 1
r−
(κ2xi∗
2
c∗2
)γ
(1 − γ )
×(
1
r
(1 − τ + τ
β
β − γ− τ
ηγ
β − γ
)− (1 − αη)
�
κ2
). (40)
uppose, on the other hand, that xi∗jN
≤ xpj
for a given debt structure j. We
btain xi∗j
= xi∗jN
and c∗j
= c∗jN
. Substituting the solutions into the equity
alue in (23) yields
oj (x) =
(x
xi∗j
)β
{Vaj
(xi∗
j , c∗j
) − I}. (41)
.2. Optimal debt structure strategies
In the previous subsection, we considered the solutions and their
alues for a given debt structure j ( j ∈ {1, 2}). In this subsection, we
xamine the solutions to the problem (22). Recall that we cannot
btain the analytical solution of Eoj(x). By comparing the magnitudes
f Eo1(x)and Eo
2(x) for given parameters, we obtain the solutions to the
roblem defined by (22) numerically.
Suppose that the basic parameters are r = 0.09, μ = 0.01, I = 5,
= 0.15, α = 0.4, and x = 0.4.16 Whether bank or market debt is
ssued depends on the combination of the three key parameters: q
debt issuance limit), σ (volatility), and η (firm’s bargaining power).
The upper left-hand side panel of Fig. 3 depicts the regions ofo1(x) > Eo
2(x) in (η,σ ) space. Three lines indicate the boundaries ofo1(x) = Eo
2(x) for q = 1, q = 0.8, and q = 0.6. Under the basic parame-
ers, the boundary for q = 1 is the same as the one for q ↑ +∞ (i.e., the
ne for the nonconstrained levered firm). For the regions of smaller σnd larger η, we see Eo
1(x) > Eo2(x), i.e., the firm prefers market debt fi-
ancing. For the regions of larger σ and smaller η, in contrast, we haveo2(x) > Eo
1(x), i.e., the firm prefers bank debt financing. Importantly,
n increase in q enlarges the regions of Eo1(x) > Eo
2(x). The next obser-
ation characterizes the effects of debt issuance limit parameter q.
bservation 1. Consider the optimization problem for the constrained
evered firm defined by (22). First, as debt issuance limit constraint q
ncreases, the firm becomes more likely to issue market debt. Second,
hen volatility σ is greater, the firm is more likely to prefer bank debt.
inally, when the firm’s bargaining power η is stronger (in other words,
hen the bank’s bargaining power is weaker), the firm is more likely to
hoose market debt.
In the upper left-hand side panels of Fig. 3, we consider three
oints: Points A, B, and C. Point A is an arbitrary case where the firm
refers market debt financing for any q. Point B is an arbitrary case
here the firm prefers market debt financing for larger q, while the
rm prefers bank debt financing for smaller q. Point C is an arbitrary
ase where the firm prefers bank debt financing for any q.
The other three panels of Fig. 3 demonstrate the equity options
alues with respect to q. The parameters of the upper right-hand,
16 Under the basic parameters, x = 0.4 always satisfies x < min{xi∗j, xi∗
jN} for all j ( j ∈
1, 2}).
h
(
Please cite this article as: T. Shibata, M. Nishihara, Investment timing
Operational Research (2014), http://dx.doi.org/10.1016/j.ejor.2014.09.011
ower left-hand, and lower right-hand side panels correspond to those
f Points A, B, and C, respectively. First, in the upper right-hand side
anel, we confirm Eo1(x) ≥ Eo
2(x) for all regions of q. Second, in the
ower left-hand side panel, we see that Eo2(x) > Eo
1(x) for q < q̂ = 0.71
nd Eo1(x) ≥ Eo
2(x) for q ≥ q̂. Finally, in the lower right-hand side panel,o2(x) ≥ Eo
1(x) for all regions of q. Consequently, we have the following
esult.
bservation 2. Consider the optimization problem for the constrained
evered firm defined by (22). For η ∈ (0, 1], there are three optimal cases,
epending on the parameters.
(i) There exists Eo2(x) ≥ Eo
1(x)for all regions of q. Then, the constrained
levered firm chooses bank debt issuance.
(ii) There exists a unique q̂ such that Eo2(x) ≥ Eo
1(x) for q ≤ q̂ and
Eo1(x) > Eo
2(x) for q > q̂. Then, the firm prefers bank debt issuance
for q ≤ q̂, while it prefers market debt issuance for q > q̂.
(iii) There exists Eo1(x) ≥ Eo
2(x) for all regions of q. Then, the firm
chooses market debt issuance.
or η = 0, on the other hand, there exists only one case. The firm chooses
ank debt issuance at the time of investment.
Observations 1 and 2 are most closely related to the stylized facts
n debt composition. According to the definition by Rajan (1992)
nd Hackbarth et al. (2007), the firms with larger q, larger η, and
maller σ (smaller q, smaller η, and larger σ ) best approximate
arge/mature (small/young) corporations. Then, based on the above
efinition, small/young firms are more likely to issue bank debt,
hereas large/mature firms are more likely to issue market debt.
n addition, these implied results in Observations 1 and 2 fit well
ith the empirical findings. Blackwell and Kidwell (1988) observe
hat small and risky firms sell debt privately. Denis and Mihov (2003)
nd Rauh and Sufi (2010) find that low-credit-quality firms (the firm
ith high volatility) are more likely to have bank debt.
. Model implications
In this section, we consider the more important implications of
ur model. Sections 4.1 and 4.2 investigate the effects of debt issuance
onstraints on the investment thresholds and coupon payments, re-
pectively. Section 4.3 shows that a symmetric relationship between
he investment thresholds and the corresponding values does not al-
ays exist. Section 4.4 considers the effects of having the choice of
ebt structure. Section 4.5 analyzes the effects of a debt issuance limit
onstraint on the credit spreads and default probabilities. Section 4.6
iscusses how the debt issuance amount is determined by debt
olders.
.1. Effects of constraints on investment thresholds
This subsection examines the effects of a debt issuance constraint
n the investment thresholds.
The upper two panels of Fig. 4 illustrate the investment thresholds
ith the debt issuance constraint ratio q (the parameters of the left-
and and right-hand side panels correspond to those of Points A and
in the upper left-hand side panel of Fig. 3, respectively). In the
eft-hand and right-hand side panels, the firm prefers market debt
ssuance and bank debt issuance, respectively. Recall from Lemma 2
hat the firm is constrained by the debt issuance bound if xpj
< xi∗jN
j ∈ {1, 2}), while it is not otherwise. In the upper left-hand side panel,
or example, if q < 0.9449 (q < 0.7045) because xp1 < xi∗
1N (xp2 < xi∗
2N),
he firm is constrained by the market (bank) debt issuance bound. We
ave limq↓0 xi∗j
= xi∗U and limq↑+∞ xi∗
j= xi∗
jNwith xi∗
U ≥ xi∗jN
. The right-
and side panel follows similarly.
Interestingly, xi∗j
is not always in between xi∗jN
and xi∗U for any q ∈
0, +∞), given a debt structure j ( j ∈ {1, 2}). This implies that the first
, debt structure, and financing constraints, European Journal of
8 T. Shibata, M. Nishihara / European Journal of Operational Research 000 (2014) 1–14
ARTICLE IN PRESSJID: EOR [m5G;October 21, 2014;15:23]
0 0.5 10
0.05
0.1
0.15
0.2
η (bargaining power)
σ (v
olat
ility
)
q=1q=0.8q=0.6
Eo1>Eo
2
Eo2>Eo
1
C
B
A
0 0.7045 0.9449
0.3057
0.4002
0.4349
q
Eo (
equi
ty o
ptio
n va
lue)
EUo
E1No
E1o
E2No
E2o
0.6 0.71 0.7945 0.92830.575
0.5928
0.6020
q
Eo (
equi
ty o
ptio
n va
lue)
E1No
E1o
E2No
E2o
0 0.9004 1.2
0.6698
0.7734
0.7869
0.8
q
Eo (
equi
ty o
ptio
n va
lue)
EUo
E1No
E1o
E2No
E2o
Fig. 3. Equity option values.
O
a
c
r
t
T
b
a
i
P
t
conjecture is no longer correct. That is, the investment thresholds xi∗j
are nonmonotonic with respect to q.
Consider the intuition of the U-shaped relationship between xi∗j
and q. When the firm comes up against the debt issuance limit con-
straints, an increase in q increases the debt value Daj(xi∗
j, c∗
j). Then there
are two conflicting effects of xi∗j
associated with increasing Daj(xi∗
j, c∗
j).
The first is the “financing constraint effect” where an increase in q
increases xi∗j
because Daj(xi∗
j, c∗
j) is increasing with respect to xi∗
j. The
second is the “leverage effect” where an increase in q decreases xi∗j
via
(x/xi∗j)β{Ea
j(xi∗
j, c∗
j)− (I − Da
j(xi∗
j, c∗
j))}. For low values of q, the second
effect dominates the first effect. For high values of q, the first effect
dominates the second effect. Thus, the investment thresholds have
a U-shaped relationship with respect to q.17 We summarize these
results as follows.
17 As demonstrated in Fig. 4, to the extent that we have solved the investment thresh-
olds numerically for various parameter values, we could not find any example of a
monotonic relationship between xi∗j
and q for any j.
a
s
t
l
t
Please cite this article as: T. Shibata, M. Nishihara, Investment timin
Operational Research (2014), http://dx.doi.org/10.1016/j.ejor.2014.09.011
bservation 3. Given a debt structure, the investment thresholds have
U-shaped relationship with the debt issuance bound q in almost all
ases.
Shibata and Nishihara (2012) have already shown a nonmonotonic
elationship between xi∗1 and q. The new result obtained in Observa-
ion 3 is that we find a nonmonotonic relationship between xi∗2 and q.
he nonmonotonicity property is consistent with theoretical studies
y Boyle and Guthrie (2003) and Hirth and Uhrig-Homburg (2010)
nd an empirical study by Cleary et al. (2007).
The two panels of Fig. 5 provide the most interesting case, assum-
ng that η = 1 and σ = 0.15 which correspond to the parameters of
oint B in the upper left-hand side of Fig. 3. Under these parame-
ers, we have already shown that the firm chooses bank debt issuance
t the time of investment for 0 < q < q̂ = 0.71, and market debt is-
uance for q ≥ q̂ = 0.71. The thick dotted line shows the investment
hresholds (coupon payments) at the equilibrium. An increase in q
eads to a jump in the investment thresholds, because the debt struc-
ure strategy has changed from bank debt to market debt. We see
g, debt structure, and financing constraints, European Journal of
T. Shibata, M. Nishihara / European Journal of Operational Research 000 (2014) 1–14 9
ARTICLE IN PRESSJID: EOR [m5G;October 21, 2014;15:23]
0 0.7045 0.9449 1.2
0.5832
0.5962
0.6404
q
xi (in
vest
men
t thr
esho
ld)
xUi∗
x1Ni∗
x1i∗
x1p
x2Ni∗
x2i∗
x2p
0 0.9317 1.2
0.7572
0.7627
0.78
0.8101
q
xi (in
vest
men
t thr
esho
ld)
xUi∗
x1Ni∗
x1i∗
x1p
x2Ni∗
x2i∗
x2p
0 0.7045 0.9449 1.20
0.3401
0.4405
0.5
q
c (c
oupo
n pa
ymen
t)
c1N∗
c1∗
c2N∗
c2∗
0 0.9004 1.20
0.4623
0.5161
0.6
q
c (c
oupo
n pa
ymen
t)
c1N∗
c1∗
c2N∗
c2∗
Fig. 4. Investment thresholds and coupon payments with q.
t
t
O
l
t
p
t
c
n
4
t
r
t
s
O
c
t
i
t
M
t
a
18 To the extent that we have solved the coupon payments numerically for various
parameter values, we could not find any example of a nonmonotonic relationship
between c∗j
and q for any j ( j ∈ {1, 2}).
hat, at q = 0.71 the investment threshold jumps from xi∗2 = 0.6612
o xi∗1 = 0.6585. We have the following result.
bservation 4. Consider the optimization problem for the constrained
evered firm defined by (22). Then, at the equilibrium, the investment
hresholds have a discontinuous W-shaped relationship with the friction
arameter q, depending on the two other key parameters, η and σ .
With the optimal choice of debt structure, the relationship be-
ween the investment thresholds and debt issuance bound is a more
omplicated nonmonotonicity relationship, compared with the sce-
ario in which firms have no choice of debt structure.
.2. Effects of constraints on coupon payments
This subsection examines the effects of a debt issuance bound on
he coupon payments.
The lower two panels of Fig. 4 show the coupon payments with
espect to q. We see that c∗j
is monotonically increasing with respect
Please cite this article as: T. Shibata, M. Nishihara, Investment timing
Operational Research (2014), http://dx.doi.org/10.1016/j.ejor.2014.09.011
o q, given the debt structure j ( j ∈ {1, 2}).18 The next observation
ummarizes this property of the coupon payments.
bservation 5. Given a debt structure, the coupon payments for the
onstrained levered firm are between those for the unlevered firm and
hose for the nonconstrained levered firm in almost all cases.
Observation 5 contrasts with Observation 3, where xi∗j
is not always
n the regions [xi∗jN
, xi∗U ]. Thus, it is less costly to distort xi∗
jaway from
he regions [xi∗jN
, xi∗U ] than to distort c∗
jaway from the regions [0, c∗
jN].
oreover, note that c∗j
≤ c∗jN
implies xd∗j
= xdj(c∗
j) ≤ xd
j(c∗
jN) = xd∗
jN.
In the right-hand side panel of Fig. 5, the thick dotted line shows
he coupon payments at the equilibrium. An increase in q leads to
jump in the coupon payments from c∗2 = 0.3510 to c∗
1 = 0.3302 at
, debt structure, and financing constraints, European Journal of
10 T. Shibata, M. Nishihara / European Journal of Operational Research 000 (2014) 1–14
ARTICLE IN PRESSJID: EOR [m5G;October 21, 2014;15:23]
0.5 0.63 0.71 0.7945 0.9283
0.6585
0.6612
0.6698
0.6725
q
xi (in
vest
men
t thr
esho
ld)
x1Ni∗
x1i∗
x2Ni∗
x2i∗
eq.
0.5 0.71 0.7945 0.9283
0.3302
0.3510
0.4146
0.4454
q
c (c
oupo
n pa
ymen
ts)
c1N∗
c1∗
c2N∗
c2∗
eq.
Fig. 5. Discontinuous solutions with respect to q.
0.1 0.1579 0.20.575
0.75
σ
x ji (in
vest
men
t thr
esho
ld)
xi∗1
xi∗2
0.1 0.1495 0.20.4
0.8
σ
Eo j (
equi
ty o
ptio
n va
lue)
Eo1
Eo2
Fig. 6. Asymmetric relationship between thresholds and values.
a
s
p
s
W
O
t
w
m
fi
19 For all the regions of σ ∈ [0.1, 0.2] with q = 0.7 and η = 1, we have xi∗jN
> xpj
( j ∈{1, 2}), implying that the firm is financially constrained by debt issuance bounds.
20 We confirm one of the results obtained in Fig. 2: namely, that an increase in σ
changes the firm’s debt capital strategies from market debt financing to bank debt
financing.
q = 0.71. The reason is that the firm changes the debt structure strate-
gies at q̂ = 0.71. We summarize the result as follows.
Observation 6. Consider the optimization problem for the constrained
levered firm defined by (22). Then, at the equilibrium, the coupon pay-
ments have a discontinuous curved relationship with the friction param-
eter q, depending on the two other key parameters, η and σ .
4.3. Relationship between investment thresholds and values
In this subsection, we consider the relationship between the in-
vestment thresholds and the corresponding values. Recall in Lemma 1
that we have a symmetric relationship (i.e., xi∗1N ≤ xi∗
2N if and only if
Eo1N(x) ≥ Eo
2N(x)) when q is sufficiently large. This result implies the
second intuitive conjecture, i.e., xi∗1 ≤ xi∗
2 if and only if Eo1(x) ≥ Eo
2(x).Fig. 6 shows the effects of volatility, σ , on the investment thresh-
olds and equity options values, respectively. The key parameters are
Please cite this article as: T. Shibata, M. Nishihara, Investment timin
Operational Research (2014), http://dx.doi.org/10.1016/j.ejor.2014.09.011
ssumed to be q = 0.7 and η = 1 for σ ∈ [0.1, 0.2].19 In the left-hand
ide panel, we have xi∗1 < xi∗
2 for σ < 0.1579. In the right-hand side
anel, we see that Eo1(x) ≥ Eo
2(x) for σ ≤ 0.1495.20 From these two re-
ults, we obtain xi∗1 < xi∗
2 and Eo1(x) < Eo
2(x) for σ ∈ (0.1495, 0.1579).e have the following result.
bservation 7. Under debt issuance limit constraints, a symmetric rela-
ionship is not always obtained. In particular, the investment thresholds
ith bank (market) debt financing are not always lower than those with
arket (bank) debt financing when the firm prefers bank (market) debt
nancing to market (bank) debt financing.
g, debt structure, and financing constraints, European Journal of
T. Shibata, M. Nishihara / European Journal of Operational Research 000 (2014) 1–14 11
ARTICLE IN PRESSJID: EOR [m5G;October 21, 2014;15:23]
i
d
s
t
4
i
w
i
i
h
o
d
c
t
(
c
o
E
l
[
d
c
e
k
O
a
o
t
t
o
m
e
e
c
4
s
s
r
d
c
k
r
a
c
T
e
fi
b
d
t
r
s
o
P
fi
c
F
d
d
c
w
i
l
d
t
a
r
l
B
fi
O
d
d
T
c
p
i
fi
f
f
t
t
fi
fi
t
O
d
o
s
p
4
e
h
o
t
d
i
t
d
d
Note that the finding in Observation 7 provides important new
nsights into the difference between the model with the choice of
ebt structure (our model) and the model without the choice of debt
tructure (the model of Shibata and Nishihara (2012)), as shown in
he next subsection.
.4. Effects of having the additional choice of debt structure
This subsection examines how debt structure choice (market debt
ssuance or bank debt issuance) influences corporate investment
hen the firm is constrained with a debt issuance bound. Recall that
t is assumed that the firm can issue only market debt at the time of
nvestment in the previous model developed by Shibata and Nishi-
ara (2012). Thus, this subsection makes clear the difference between
ur model and the previous model.
Suppose, as a benchmark, that the firm is not constrained by a
ebt issuance bound (i.e., q is sufficiently large). Then, having the
hoice of market debt issuance or bank debt issuance always increases
he equity options values and always hastens corporate investment
decreases the investment threshold), compared with having only the
hoice of market debt issuance. The reason is a straightforward result
f the symmetric relationship (i.e., xi∗1N ≥ xi∗
2N if and only if Eo1N(x) ≤
o2N(x)).
Suppose that the firm comes up against the debt issuance
imit constraint (for example, assume q = 0.7, η = 1, and σ ∈0.1495, 0.1579]). Then, interestingly, having the choice of market
ebt issuance or bank debt issuance delays corporate investment (in-
reases the investment threshold) although it always increases the
quity options value, compared with having only the choice of mar-
et debt issuance. Thus, we summarize these results as follows.
bservation 8. Suppose that the firm faces the debt issuance constraint
t the time of investment. Then, having the choice of market debt issuance
r bank debt issuance does not always hasten corporate investment al-
hough it always increases the equity options value.
Our model is an extension of Shibata and Nishihara (2012) in that
he firm can choose the optimal debt structure (i.e., the firm has the
ption of issuing bank debt). Our intuition is that choosing the opti-
al debt structure enables the firm to hasten its corporate investment
ven though the firm is constrained by the debt issuance bound. How-
ver, interestingly, this intuition is not always correct when the firm
omes up against financial constraints.
.5. Risk and return for debt holders
In this subsection, we consider the effects of debt issuance con-
traints on debt holders. To do so, the two key variables are credit
preads and default probabilities, which are regarded as measures of
eturn and risk for debt holders, respectively. The credit spread and
efault probability are defined as
sk := c∗k
Daj(xi∗
k, c∗
k)
− r, pk := ET i∗k
[e−r(T i∗k
−Td∗k
)] =(
xi∗k
xdj(c∗
k)
)γ
,
∈ {j, jN}, (42)
espectively (j ∈ {1, 2}). Note that the two measures above are defined
t the optimal investment time T i∗k
.
We have already recognized that debt issuance constraints cause
∗j ≤ c∗
jN, xd∗j ≤ xd∗
jN , Daj
(xi∗
j , c∗j
) ≤ Daj
(xi∗
jN, c∗jN
),
j ∈ {1, 2}. (43)
he first inequality has already been shown in the lower two pan-
ls of Fig. 4; the second inequality is straightforward because of the
rst inequality and xd∗k
= xdj(c∗
k); and the third inequality is obvious
Please cite this article as: T. Shibata, M. Nishihara, Investment timing
Operational Research (2014), http://dx.doi.org/10.1016/j.ejor.2014.09.011
ecause of the definition of debt issuance constraints. Moreover, the
ebt issuance constraint does not always cause xi∗j
≥ xi∗jN
, as shown in
he upper two panels of Fig. 4. Thus, using these previously presented
esults, we cannot determine which is bigger: csj or csjN and pj or pjN.
The upper two and lower two panels of Fig. 7 depict the credit
preads and the default probabilities, respectively. The parameters
f the left-hand and right-hand side panels correspond to those of
oints A and C in the upper left-hand side panel of Fig. 3, where the
rm prefers market and bank debt financing, respectively. We see
sj ≤ csjN, pj ≤ pjN. (44)
or both market and bank debt financing, we have the result that the
ebt issuance limit constraint causes a decrease in credit spreads and
efault probabilities. Regarding the first inequality, the debt issuance
onstraint leads to a decrease in the ratio of c∗j
to Daj(xi∗
j, c∗
j), compared
ith the ratio of c∗jN
to DajN
(xi∗jN
, c∗jN
). The implication of the second
nequality is that, because γ < 0, the distance between xi∗j
and xd∗j
is
arger than the distance between xi∗jN
and xd∗jN
. This implies that the
efault for the constrained levered firm is less likely to occur than
hat for the unconstrained levered firm.
Thus, the more severe the debt issuance bounds are, the lower
re the credit spreads and default probabilities. These links between
isk and return are the same as those traditionally suggested in the
iterature and also match the results by Gomes and Schmid (2010) and
orgonovo and Gatti (2013). The next characterizes the properties of
nancing constraints.
bservation 9. Debt financing constraints lead to a low-risk (low-
efault probabilities) and low-return (low-credit spreads) situation for
ebt holders.
In addition, in both panels of Fig. 7, we have cs2 ≥ cs1 and p2 ≥ p1.
hese inequalities imply that bank debt is high risk and high return,
ompared with market debt. Importantly, irrespective of the firm’s
reference of debt structure, the credit spread and default probabil-
ties for bank debt financing are larger than those for market debt
nancing. Regarding the first inequality, the ratio of c∗2 to Da
2(xi∗2 , c∗
2)
or bank debt issuance is higher than the ratio of c∗1 to Da
1(xi∗1 , c∗
1)or market debt issuance. The implication of the second inequality is
hat, because of γ < 0, the distance between xi∗2 and xd∗
2 is smaller
han the distance between xi∗1 and xd∗
1 , implying that default of the
rm financed by bank debt is more likely to occur than default of the
rm financed by market debt. These numerical simulations suggest
he following observation.
bservation 10. The credit spreads and default probabilities for bank
ebt are larger than those for market debt in almost all cases.
The results presented are consistent with the empirical results
f Davydenko and Strebulaev (2007) who find that the possibility of
trategic debt service increases corporate credit spreads and default
robabilities.
.6. Amount of debt issuance
So far, we have assumed that the debt issuance amount (i.e., q) is
xogenous. This subsection examines how q is determined by debt
olders for a given debt structure j ( j ∈ {1, 2}).
Note that there is no arbitrage opportunity for debt issuance in
ur model, where debt holders provide the amount of qI and take
he value of Daj(xi∗
j, c∗
j) = qI. Additionally, credit spreads (return) and
efault probabilities (risk) are monotonically increasing with q. This
mplies that a decrease in q will enable debt holders to decrease
heir return and risk. In practice, it is natural for debt holders to
etermine q assuming, e.g., credit spreads are lower than 20 bp or
efault probabilities are lower than 5 percent. We now suppose that
, debt structure, and financing constraints, European Journal of
12 T. Shibata, M. Nishihara / European Journal of Operational Research 000 (2014) 1–14
ARTICLE IN PRESSJID: EOR [m5G;October 21, 2014;15:23]
0 0.57 0.65 0.85 1.20
20
32.5119
40
65.6535
q
cs (
cred
it sp
read
, bp)
cs1N
cs1
cs2N
cs2
0 0.33 0.46 0.63 0.90 1.20
20
40
92.3965
246.4458
q
cs (
cred
it sp
read
, bp)
cs1N
cs1
cs2N
cs2
0 0.48 0.55 0.90 1.20
56.0283
10
39.2431
q
p (d
efau
lt pr
obab
ility
, %)
p1N
p1
p2N
p2
0 0.30 0.42 0.56 0.79 0.90 1.20
5
10
13.9647
62.0389
70
q
p (d
efau
lt pr
obab
ility
, %)
p1N
p1
p2N
p2
Fig. 7. Low-risk and low-return of debt holders by financing constraints.
t
u
a
r
a
f
5
k
5
s
d
s
fi
a
c
t
q is determined on the basis of amount of return and risk debt holders
assume.
In the upper left-hand side panel of Fig. 7 (σ = 0.1), suppose
that q is decided such that credit spreads are lower than 20 bp. We
then have q = 0.5759 and q = 0.8564 for the bank debt and mar-
ket debt, respectively, which gives the result Da2(x
i∗2 , c∗
2) = 2.8795 and
Da1(x
i∗1 , c∗
1) = 4.2820. Additionally, suppose that credit spreads are al-
lowed up to 40 bp. We then have q = 0.6486 (i.e., Da2(x
i∗2 , c∗
2) = 3.2430)
and q = 0.9449 (i.e., Da1(x
i∗1 , c∗
1) = 4.7245). By increasing the upper
limit constraints of credit spreads, q is increased. In the upper right-
hand side panel (σ = 0.2), for the upper limit constraints of credit
spreads allowed up to 20 bp, we have q = 0.3385 and q = 0.4648 for
bank and market debt, respectively. Thus, an increase in σ decreases
q (i.e., the debt issuance amount).
Similarly, we examine the upper limit constraints of default prob-
abilities. In the lower left-hand side panel of Fig. 7 (σ = 0.1), sup-
pose that q is decided such that the default probability is lower than
5 percent. We then have q = 0.4856 (i.e., Da2(x
i∗2 , c∗
2) = 2.4280) and q =0.9077 (i.e., Da(xi∗, c∗) = 4.5385) for bank and market debt, respec-
1 1 1Please cite this article as: T. Shibata, M. Nishihara, Investment timin
Operational Research (2014), http://dx.doi.org/10.1016/j.ejor.2014.09.011
ively. Additionally, suppose that default probabilities are allowed
p to 10 percent. We then have q = 0.5525 (i.e., Da2(x
i∗2 , c∗
2) = 2.7625)
nd q = 0.9449 (i.e., Da1(x
i∗1 , c∗
1) = 4.7245) for bank and market debt,
espectively. By increasing the upper limit constraints of default prob-
bilities, q is increased. In the lower right-hand side panel (σ = 0.2),
or the upper limit constraints of default probabilities allowed up to
percent, we then have q = 0.3031 and q = 0.5688 for bank and mar-
et debt, respectively. Thus, an increase in σ decreases q.
. Concluding remarks
We have proposed a model that analyzes optimal investment
trategies under an optimal structure of bank debt and market
ebt with issuance limit constraints. By considering various debt
tructures, we shed light on decisions regarding a firm’s investment,
nancing, and debt choice strategies. As a result, we can discuss the
vailability and effects of various debt structures under issuance
onstraints.
We have four important results with respect to financial fric-
ions. The first result is which type of debt is issued at the time of
g, debt structure, and financing constraints, European Journal of
T. Shibata, M. Nishihara / European Journal of Operational Research 000 (2014) 1–14 13
ARTICLE IN PRESSJID: EOR [m5G;October 21, 2014;15:23]
i
v
d
fi
t
l
o
m
b
s
n
b
c
O
t
T
c
a
i
h
t
c
s
s
t
t
u
t
a
T
d
d
o
u
f
a
s
a
p
W
c
A
e
t
(
C
v
U
v
c
I
K
J
m
A
P
(
t
s
s
L
w
K
a
A
F
F
s
T
s
a
R
B
B
B
B
B
B
B
B
nvestment depends largely on three parameters (debt issuance limit,
olatility, and bargaining power in the renegotiation during financial
istress). In particular, increasing the debt issuance limit makes a
rm more likely to issue market debt than bank debt. In practice,
he firms with higher (lower) debt issuance bounds are regarded as
arge/mature (small/young) corporations. Based on this definition,
ur results show that large/mature (small/young) corporations are
ore likely to choose market (bank) debt. Thus, this segmentation is
roadly consistent with the stylized facts.
The second result is that, for a given debt structure, the relation-
hip between investment thresholds and the debt issuance limit is
onmonotonic. Because we find a nonmonotonic relationship under
ank debt financing in addition to market debt financing, we can
onclude that the nonmonotonic relationship is theoretically robust.
n the other hand, for a given debt structure, the relationship be-
ween coupon payments and the debt issuance limit is monotonic.
hus, it is less costly to distort investment thresholds than to distort
oupon payments. Moreover, if the optimal debt structure strategies
re changed from bank debt to market debt by increasing the debt
ssuance bound, the investment thresholds (and coupon payments)
ave a discontinuous relationship with the debt issuance bound. Thus,
he optimal debt structure makes the corporate investment (and
oupon payments) strategy more complicated, compared with the
cenario in which firms have no choice of debt structure.
The third result is that there is not always a symmetric relation-
hip between investment thresholds and equity option values when
here is a debt issuance constraint. More precisely, the investment
hresholds under bank debt financing are not always lower than those
nder market debt financing, even when the firm prefers bank debt
o market debt. This result contrasts with the result where there is
symmetric relationship when there is no debt issuance constraint.
hus, under a debt issuance constraint, having the choice of market
ebt or bank debt (having the additional choice of debt structure)
oes not always accelerate corporate investment although its equity
ption value is increased. We conclude that investment strategies
nder financing frictions are different from those under no financing
rictions.
The final result is that debt issuance constraints create low-risk
nd low-return situation for debt holders. More precisely, the more
evere the debt issuance limits are, the lower are the credit spreads
nd default probabilities. In addition, the credit spreads and default
robabilities for bank debt are larger than those for market debt.
e conclude that bank debt is relatively high risk and high return,
ompared with market debt.
cknowledgments
We thank seminar participants at the Advanced Methods in Math-
matical Finance Conference (Angers), the 7th World Congress of
he Bachelier Finance Society (Sydney), the EURO XXVI Conference
Rome), the 12th SAET Conference (Brisbane), the Bank of Japan, the
hinese University of Hong Kong, the Fields Institute, Hokkaido Uni-
ersity, the Hong Kong University of Science and Technology, Kyoto
niversity, Nanzan University, National University of Singapore, Uni-
ersity College London, and the University of Tokyo for their helpful
omments. This research was supported by a Grant-in-Aid from the
shii Memorial Securities Research Promotion Foundation, the JSPS
AKENHI (Grant Numbers 26242028, 26285071, and 26350424), the
SPS Postdoctoral Fellowships for Research Abroad, and the Telecom-
unications Advances Foundation.
ppendix A
In this subsection, we derive the functions fj1, fj2, fj3, and fj4 in
roposition 1. Note that fjk is a function of (xi∗j
, c∗j) for all j and k
j ∈ {1, 2}, k ∈ {1, 2, 3, 4}). Because the derivations of f and f are
1k 2kPlease cite this article as: T. Shibata, M. Nishihara, Investment timing
Operational Research (2014), http://dx.doi.org/10.1016/j.ejor.2014.09.011
he same (k ∈ {1, 2, 3, 4}), we show only the derivation of f2k. In the
ubgame to the investment decision problem for the firm with con-
traint bank debt financing, the Lagrangian is formulated as
= xi2
−β(�xi
2 + τ c2
r− τ c2
r
β
β − γ
(κ2xi
2
c2
)γ
− I
)
+λ
(qI − c2
r− c2
((1 − αη)
�
κ2
− 1
r
(1 − τ + τ
β
β − γ− τ
ηγ
β − γ
))(κ2xi
2
c2
)γ )), (A.1)
here λ ≥ 0 denotes the multiplier on the constraint. The Karush–
uhn–Tucker conditions are given by
∂L∂xi
2
= −βxi∗2
−β−1(�xi
2 + τ c2
r− τ c2
r
β
β − γ
(κ2xi
2
c2
)γ
− I
)
+ xi∗2
−β(� − τ c∗
2
r
β
β − γ
(κ2xi∗
2
c∗2
)γ
γ xi∗2
−1)
−λc∗2
((1 − αη)
�
κ2− 1
r
(1 − τ + τ
β
β − γ− τ
ηγ
β − γ
))
×(
κ2xi∗2
c∗2
)γ
γ xi∗2
−1 = 0, (A.2)
∂L∂c2
= xi∗2
−β(
τ
r− τ
r
β
β − γ
(κ2xi∗
2
c∗2
)γ
(1 − γ )
)
−λ
(1
r+
((1 − αη)
�
κ2− 1
r
(1 − τ + τ
β
β − γ− τ
ηγ
β − γ
))
× (1 − γ )
(κ2xi∗
2
c∗2
)γ )= 0, (A.3)
nd
λ
(qI − c∗
2
r− c∗
2
((1 − αη)
�
κ2− 1
r
(1 − τ + τ
β
β − γ− τ
ηγ
β − γ
))
×(
κ2xi∗2
c∗2
)γ ))= 0. (A.4)
ssume that λ > 0, which is equivalent to xi∗2 > x1 from Lemma 2.
or q = 0, because c∗2 = 0 from (A.4), we have xi∗
2 = xi∗U and xd∗
2 = 0.
or q > 0, we have c∗2 > 0. Then (xi∗
2 , c∗2)are obtained uniquely by two
imultaneous equations:
f21(xi∗2 , c∗
2)
f22(xi∗2 , c∗
2)− f23(xi∗
2 , c∗2)
f24(xi∗2 , c∗
2)= 0, Da
2(xi∗2 , c∗
2)− qI = 0. (A.5)
he first equation is given by rearranging (A.2) and (A.3), while the
econd equation is given by (A.4). Thus, f21, f22, f23, and f24 are obtained
s (37)–(40). Similarly, we can derive f11, f12, f13, and f14.
eferences
lack, F., & Cox, J. C. (1976). Valuing corporate securities: Some effects of bond indenture
provisions. Journal of Finance, 31(2), 351–367.lackwell, D. W., & Kidwell, D. S. (1988). An investigation of cost differences between
public sales and private placements of debt. Journal of Financial Economics, 22(2),
253–278.olton, P., & Freixas, X. (2000). Equity, bonds, and bank debt: Capital structure and finan-
cial market equilibrium under asymmetric information. Journal of Political Economy,108(2), 324–351.
orgonovo, E., & Gatti, S. (2013). Risk analysis with contractual default. Does covenantbreach matter? European Journal of Operational Research, 230(2), 431–443.
oyle, G. W., & Guthrie, G. A. (2003). Investment, uncertainty, and liquidity. Journal ofFinance, 63(5), 2143–2166.
rennan, A., & Schwartz, E. (1984). Optimal financial policy and firm valuation. Journal
of Finance, 39(3), 593–607.roadie, M., Chernov, M., & Sundaresan, S. (2007). Optimal debt and equity values in
the presence of Chapter 7 and Chapter 11. Journal of Finance, 62(3), 1341–1377.ulow, J. I., & Shoven, J. B. (1978). The bankruptcy decision. Bell Journal of Economics,
9(2), 436–445.
, debt structure, and financing constraints, European Journal of
14 T. Shibata, M. Nishihara / European Journal of Operational Research 000 (2014) 1–14
ARTICLE IN PRESSJID: EOR [m5G;October 21, 2014;15:23]
L
L
L
M
M
M
M
M
N
N
R
W
Cleary, S., Povel, P., & Raith, M. (2007). The U-shaped investment curve: Theory andevidence. Journal of Financial and Quantitative Analysis, 42(1), 1–40.
Davydenko, S. A., & Strebulaev, I. A. (2007). Strategic actions and credit spreads: Anempirical investigation. Journal of Finance, 62(6), 2633–2671.
Denis, D., & Mihov, V. (2003). The choice among bank debt, non-bank private debt,and public debt: Evidence from new corporate borrowings. Journal of Financial
Economics, 70(1), 3–28.Dixit, A. (1989). Entry and exit decisions under uncertainty. Journal of Political Economy,
97(3), 620–638.
Dixit, A., & Pindyck, R. S. (1994). Investment under uncertainty. Princeton, NJ: PrincetonUniversity Press.
Fan, H., & Sundaresan, S. (2000). Debt valuation, strategic debt service and optimaldividend policy. Review of Financial Studies, 13(4), 1057–1099.
Fazzari, S. M., Hubbard, R. G., & Petersen, B. C. (1988). Financing constraints and corpo-rate investment. Brookings Papers on Economic Activity, 1988(1), 141–206.
Gale, D., & Hellwig, H. (1985). Incentive compatible debt contracts: The one period
problem. Review of Economic Studies, 52(4), 647–664.Gertner, R. H., & Scharfstein, D. S. (1991). A theory of workouts and the effect of reor-
ganization law. Journal of Finance, 46(4), 1189–1222.Gilson, S. C., John, K., & Lang, L. (1990). Troubled debt restructurings: An empirical study
of private reorganization of firms in default. Journal of Financial Economics, 27(2),315–353.
Gomes, J. F., & Schmid, L. (2010). Levered returns. Journal of Finance, 65(2), 467–494.
Greenwald, B., & Stiglitz, J. E. (1993). Financial market imperfections and businesscycles. Quaterly Journal of Economics, 108(1), 77–114.
Hackbarth, D., Hennessy, C. A., & Leland, H. E. (2007). Can the trade-off theory explaindebt structure? Review of Financial Studies, 20(5), 1389–1428.
Hennessy, C. A., & Whited, T. M. (2007). How costly is external financing? Evidencefrom a structural estimation. Journal of Finance, 62(4), 1705–1745.
Hirth, S., & Uhrig-Homburg, M. (2010). Investment timing, liquidity, and agency costs
of debt. Journal of Corporate Finance, 16(2), 243–258.Hoshi, T., Kashyap, A., & Scharfstein, D. (1991). Corporate structure, liquidity, and in-
vestment: Evidence from Japanese industrial groups. Quarterly Journal of Economics,106(1), 33–60.
Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agencycosts and ownership structure. Journal of Financial Economics, 3(4), 305–360.
Kaplan, S. N., & Zingales, L. (1997). Do investment-cash flow sensitivities provide useful
measures of financing constraints? Quaterly Journal of Economics, 112(1), 169–215.Kort, P. M., Murto, P., & Pawlina, G. (2010). Uncertainty and stepwise investment.
European Journal of Operational Research, 202(1), 196–203.
Please cite this article as: T. Shibata, M. Nishihara, Investment timin
Operational Research (2014), http://dx.doi.org/10.1016/j.ejor.2014.09.011
eland, H. E. (1994). Corporate debt value, bond covenants, and optimal capital struc-ture. Journal of Finance, 49(4), 1213–1259.
ivdan, D., Sapriza, H., & Zhang, L. (2009). Financially constrained stock returns. Journalof Finance, 64(4), 1827–1862.
ummer, S. L., & McConnell, J. J. (1989). Further evidence on the bank lending pro-cess and the capital-market response to bank loan agreements. Journal of Financial
Economics, 25(1), 99–122.auer, D. C., & Triantis, A. (1994). Interactions of corporate financing and investment
decisions. Journal of Finance, 49(4), 1253–1277.
cDonald, R., & Siegel, D. R. (1986). The value of waiting to invest. Quarterly Journal ofEconomics, 101(4), 707–727.
ella-Barral, P., & Perraudin, W. (1997). Strategic debt service. Journal of Finance, 52(2),531–556.
odigliani, F., & Miller, M. H. (1958). The cost of capital, corporation finance, and thetheory of investment. American Economic Review, 48(3), 261–297.
yers, S. C. (1977). Determinants of corporate borrowing. Journal of Financial Economics,
5(2), 147–175.ishihara, M., & Shibata, T. (2010). Interactions between preemptive competition and
a financing constraint. Journal of Economics and Management Strategy, 19(4), 1013–1042.
ishihara, M., & Shibata, T. (2013). The effects of external financing costs on investmenttiming and sizing decisions. Journal of Banking and Finance, 37(4), 1160–1175.
ajan, R. (1992). Insiders and outsiders: The choice between informed and arm’s length
debt. Journal of Finance, 47(4), 1367–1400.Rauh, J., & Sufi, A. (2010). Capital structure and debt structure. Review of Financial
Studies, 23(12), 4242–4280.Shibata, T., & Nishihara, M. (2010). Dynamic investment and capital structure under
manager-shareholder conflict. Journal of Economic Dynamics and Control, 34(2), 158–178.
Shibata, T., & Nishihara, M. (2011). Interactions between investment timing and man-
agement effort under asymmetric information: Costs and benefits of privatizedfirm. European Journal of Operational Research, 215(3), 688–696.
Shibata, T., & Nishihara, M. (2012). Investment timing under debt issuance constraint.Journal of Banking and Finance, 36(4), 981–991.
Stiglitz, J. E., & Weiss, A. (1981). Credit rationing in markets with imperfect information.American Economic Review, 71(3), 393–410.
Sundaresan, S., & Wang, N. (2007). Investment under uncertainty with strategic debt
service. American Economic Review Paper and Proceedings, 97(2), 256–261.ong, K. P. (2010). On the neutrality of debt in investment intensity. Annal of Finance,
6(6), 335–356.
g, debt structure, and financing constraints, European Journal of